Spanish, and a test that the next language cannot ship unreachable - #469
Merged
Conversation
Adds Spanish (es) to the language table, covering all 461 English keys. Neutral Spanish throughout, no regional forms, so it reads the same in Spain and across Latin America. Domain vocabulary was chosen by a working musician rather than translated literally, and two English pairs that collapse into one Spanish word were separated deliberately: - track -> "pista" and stems -> "stems", so the 60 keys using one and the 53 using the other stay distinguishable - key -> "tonalidad" and pitch -> "tono", which both read as "tono" otherwise Mute and Solo stay untranslated as controls, matching how they are spoken in a studio, while a full sentence uses "silenciado". GRID keeps its English label: the neutral Spanish word is long enough to overflow the beatgrid toolbar, and the regional alternative would not read as Spanish everywhere. Five strings were shortened after measuring growth against English, so the settings tabs and the beatgrid toolbar do not overflow. Only the one/other plural forms are defined, as Spanish has no few/many and plural() only routes those for Polish. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The Spanish table was complete against the English it branched from. It landed four keys short only because settings.autoDelete.* went into main afterwards, in #464. Written to match the conventions already in the table: pista for track, eliminar for delete, "no se puede deshacer" from settings.resetData.desc, "salvo que" from settings.cookies.desc, and the tu form used throughout. _detectDefault also had no branch for es, so a Spanish browser would have been handed English and the table would only ever have been found by someone who went looking in Settings. One table covers every es-* locale, deliberately, because the vocabulary carries no regional forms. Note for a separate issue: fr has this same gap and always has. A French browser gets English today. Left alone here rather than fixed in passing, since it is a different language's bug and predates this branch. Verified on the merged file rather than in review: all ten tables complete at 443 base keys, no extras or duplicates; placeholder and HTML-tag parity across every Spanish string; plural families one/other matching de and pt; node --check and the four tests/js suites; the 46 browser tests; and an overflow audit in Spanish across all five settings tabs, which came back clean including the four strings added here.
_detectDefault had branches for pl, ja, zh, de, pt-PT, pt and id, and none for fr. A browser reporting fr-FR fell through to English. The French table has been complete for some time and appears in the picker; it simply was never chosen on its own, so a French speaker had to find the setting and pick it by hand. The missing line is the smaller half. LANGUAGES, TRANSLATIONS and _detectDefault are three lists that have to agree, and only the first two are checked by anything. The failure mode is silent by construction: nothing throws, no warning appears, the picker looks right, and every string is translated the moment the language is selected. The only symptom is that it never is. tests/js/i18n-detect.test.mjs now drives _detectDefault with the locales a speaker of each shipped language actually sends, and fails if any of them lands somewhere else. It also pins the parts that are easy to break while editing: pt-PT ahead of pt, case-insensitive matching, and an unknown tag falling back to English rather than throwing. Confirmed the test can fail rather than only pass: removing the fr line again takes it from 42 passed to 39 passed, 3 failed, naming fr, fr-FR and fr-CA. _detectDefault is exported for this. It is a pure function of navigator and has no other caller.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #468
Closes #467
Spanish, contributed by @diegoarusso and offered in #447. He could not open this himself: the repo restricts pull requests to collaborators, which he flagged in that thread. The commit is his, authorship intact, pulled from his fork.
The contribution
One table, 443 keys, plus the
LANGUAGESandTRANSLATIONSentries. Nothing else. I reviewed the whole diff rather than the description, since it came from outside: it is data, with no executable code anywhere in it.one/other, 15 families, matchingdeandpt?or!without its opening¿/¡His "neutral Spanish" claim holds under measurement rather than assertion:
archivoseventeen times andficheronever, noordenadororcomputadoraat all, and the tú form in eleven places with usted in none.The domain vocabulary is the part that would have been hardest to get from a translator rather than a musician.
tiempo fuertefor downbeat,compásfor meter,golpe de bateríafor drum hit, andtonalidadagainsttonoto keep key and pitch apart where Spanish wants to collapse them.What I added on top
Four keys.
settings.autoDelete.*landed in #464 after his branch was cut, so his table was complete against the English it left. Written to his conventions:pista,eliminar, "no se puede deshacer" from hissettings.resetData.desc, "salvo que" from hissettings.cookies.desc.Two lines in
_detectDefault, and this is the part worth reading.Spanish had no branch there, so the table would have shipped unreachable: present in the picker, complete, and never chosen by anyone whose browser said
es-MX. While confirming that, French turned out to have the same gap and to have shipped with it. That is #467.A test, because the line was never the real problem.
LANGUAGES,TRANSLATIONSand_detectDefaultare three lists that must agree and only the first two were checked by anything. The failure is silent by construction: nothing throws, no warning appears, the picker looks right, and every string is correct the moment the language is chosen. The only symptom is that it never is.tests/js/i18n-detect.test.mjsdrives_detectDefaultwith the locales a speaker of each shipped language actually sends, and pins what is easy to break while editing that function:pt-PTahead ofpt, case-insensitive matching, and an unknown tag falling back to English rather than throwing. It runs in CI already, sincejs-syntaxexecutes everytests/js/*.test.mjs.I confirmed it can fail rather than only pass. Removing the French line again takes it from 42 passed to 39 passed and 3 failed, naming
fr,fr-FRandfr-CA.Verified
All ten tables complete at 443 base keys. Every
static/js/*.jspassesnode --check. All fivetests/jssuites pass. The 46 browser tests pass. The Spanish overflow audit covers all five settings tabs including the four strings added here, which were the only ones nobody had looked at.Worth a separate look
The setting that blocked this PR is still in place, and
CONTRIBUTING.mdinvites outside pull requests. Anyone who is not already a collaborator hits the same wall.